Search CTRL + K

Lecture 17 - Global Gauss Bonnet Theorem

Theorem

Let RS be a regular region of an oriented surface and let δR be made up of n closed piecewise simple regular curves, c1,...,cn. Then:

inφi+RK dA+δRKgds=2πχ

Where χ is the Euler characteristic of the region and φi are the external angles formed by the intersection of the curves.

Sketch of the proof

Let RS be a regular region of an oriented surface S. Let R be the boundary of R be made up by closed piecewise simple regular curves c1,...,cn, and let φi be the exterior angle formed by the intersection of these curves.
Let T={Ri:i=1,...,F}, where F is the number of faces, be a triangulation of R for each i. Let eij,j=1,2,3 be the edges and let ϑij denote the exterior angle formed by the triangulation and θij denote the respective interior angle with the relation θij=πϑij.
Using the local Gauss Bonett Theorem on each face we obtain:

i=1F(KdA+j=13γijKgds+j=13(πθij))=i=1F2π

The edges shared by faces are cancelled out during the integration:

RKdA+δRKgds+i,j(πθij)=i=1F2π

I will now introduce the following notation:

  • Ee= number of external edges of T
  • Ei = the number of internal edges of T
  • Ve = the number of external vertices of T
  • Vi = the number of internal vertices of T

As an example let us look at the following triangulation of a region:
Triangulation.png|500
We have Ee=3,Ei=3,Ve=3,Vi=1,F=3. We convince ourselves of the following relations:

3F=2Ei+EeEe=Ve

Thus ij(πθij)=3πFijθij=2πEi+πEeijθij.
The sum of the all the internal angles around an internal vertex is 2π and if a vertex was introduced such that it lies on a curve (thus it is an external vertex) then the internal angles add up to π. For clarity we will split Ve=Vet+Vec where Vet denotes the vertices added during triangulation while Vec are the 'natural' vertices formed by the curves ci making up the boundary of the region. Naturally we have that the sum of the interior angles corresponding to Vec is equal to i=1n(πφi) Thus we have that:

i,jθij=2πVi+πVet+i=1n(πφi)=2πVi+πVet+πVeck=1n(φk)=2πVi+πVek=1n(φk)

Now using the fact Ee=Ve and adding and subtracting πEe we obtain that:

ij(πθij)=2πE2πV+i=1nφi

Thus finally:

RKdA+δRKgds+i=1nφi=2π(FE+V)=2πχ
Corollary

For a sphere S if we take the boundary of our region R to be a great circle then:

RK=2πχ

Consider a smooth vector field F(x,y)=(P(x,y),Q(x,y)) on a subset U of R2. Let C(t)=(x(t),y(t)) be a parameterized curve in U from a point p to q with atb.
A necessary condition for the vector field to be the gradient to some scalar function f is Py=Qx.

We are interested in asking if PyQx=0 holds for a vector field F=(P,Q) on U then is it the gradient of some scalar function f(x,y) on U?

Vector Field Differential Form
F=(P,Q) ω=Pdx+Qdy
Grad f=(fx,fy) df=fxdx+fydy
QxPy=0 dω=(QxPy)dxdy=0
If ω=df this is called the exact form.

Let Zk(M) be the vector space of all closed k-forms and Bk(M) be the vector space of all exact kforms. Then:

Hk(M)=Zk(M)/Bk(M)

measures the extent to which closed kforms fail to be exact.